>How about if they can read your raw tape drive?
>
>I.e. I have just reproduced this using /dev/rst0, and have a dump should
>anyone care to see it.
>
>Interestingly enough, the crash happens during or after the I/O -- using a
>tape drive makes this painfully obvious :-).

Hmm.. once upon a time I happened to use my tape drive when I was
running X11. Suddenly the whole screen went artistic (sort of like
every second vertical line garbage). The same happened when I was
testing the throughput difference between block and character devices
for disk partitions with dd (amazing how big the difference was - 3M
from char, 1M from block, for IBM "Spitfire"). I have not reported
these because my NetBSD is not straight from the distribution
(pre-1.1-current Amiga with ISA framework from Niklas Hallqvist and
if_ed and grf_cl munged by myself). I thought that my local changes
might be the cause.

But anyway - this might suggest that the problem has been there before
but now the kernel catches the problem and panics.